The American dog tick, Dermacentor variabilis (Say) (Acari: Ixodidae), is a vector for several human disease-causing pathogens such as tularemia, Rocky Mountain spotted fever, and the understudied spotted fever group rickettsiae (SFGR) infection caused by Rickettsia montanensis. It is important for public health planning and intervention to understand the distribution of this tick and pathogen encounter risk. Risk is often described in terms of vector distribution, but greatest risk may be concentrated where more vectors are positive for a given pathogen. When assessing species distributions, the choice of modeling framework and spatial layers used to make predictions are important. We first updated the modeled distribution of D. variabilis and R. montanensis using maximum entropy (MaxEnt), refining bioclimatic data inputs, and including soil variables. We then compared geospatial predictions from five species distribution modeling frameworks. In contrast to previous work, we additionally assessed whether the R. montanensis positive D. variabilis distribution is nested within a larger overall D. variabilis distribution, representing a fitness cost hypothesis. We found that 1) adding soil layers improved the accuracy of the MaxEnt model; 2) the predicted ‘infected niche’ was smaller than the overall predicted niche across all models; and 3) each model predicted different sizes of suitable niche, at different levels of probability. Importantly, the models were not directly comparable in output style, which could create confusion in interpretation when developing planning tools. The random forest (RF) model had the best measured validity and fit, suggesting it may be most appropriate to these data.